Dynamic value alignment through preference aggregation of multiple objectives
Abstract: The development of ethical AI systems is currently geared toward setting objective functions that align with human objectives.
However, finding such functions remains a research challenge, while in RL, setting rewards by hand is a fairly standard approach.
We present a methodology for dynamic value alignment, where the values that are to be aligned with are dynamically changing, using a multiple-objective approach. We apply this approach to extend Deep Q-Learning to accommodate multiple objectives and evaluate this method on a simplified two-leg intersection controlled by a switching agent. Our approach dynamically accommodates the preferences of drivers on the system and achieves better overall performance across three metrics (speeds, stops, and waits) while integrating objectives that have competing or conflicting actions.
Submission Length: Regular submission (no more than 12 pages of main content)
Changes Since Last Submission: The changes indicated by the answer to our answer of reviewer 8qqZ are included. The current and previous changes are indicated in blue and the new changes have been indicated to the reviewer.
Assigned Action Editor: ~Aleksandra_Faust1
Submission Number: 1126
Loading